At the same Cloud Next 2026, Google didn't just announce the TPU 8 — it also launched Workspace Intelligence.
Hardware makes headlines, but software is where the daily revenue comes from.
Not a new feature, but a semantic layer
The core design of Workspace Intelligence: build a shared semantic layer across Gmail, Docs, Sheets, Slides, Chat, and Meet.
Gemini knows your entire work context — emails, calendar, files, meeting notes — and doesn't answer questions in isolation, but works as an "AI colleague who knows your full work state."
In plain English: previously, when you asked Gemini to draft an email, it didn't know your history with the recipient and could only rely on what you pasted in. Now it can directly scan your Gmail and Drive, drafting with full context.
How each product changes
Gmail: AI Inbox + AI Overviews. Gemini scans multiple threads, provides summaries and insights, and drafts replies directly in the inbox.
Google Chat: A unified command line. A single command like "Schedule a meeting for 2 PM tomorrow and attach the relevant documents" — Gemini executes it without you needing to switch between apps.
Docs / Slides: Pulls relevant content from email, chat, and Drive to draft professional documents and presentations, rendered according to company templates and branding.
Sheets: New Sheets Canvas imports external data from sources like Salesforce and HubSpot to generate interactive dashboards, without needing to write connectors yourself.
Meet: The "Take Notes for Me" feature is expanded — previously it only worked with Google Meet, but now it generates summaries and action items for meetings held on Zoom or Teams (as long as you use a Google device), supports 24 languages, and offers customizable digital avatars.
Skills: building blocks for automated workflows
Google also introduced a "Skills" mechanism, allowing teams to build their own automated workflows.
The official example: an agent that automatically cross-checks incoming invoices against historical documents.
This logic is nearly identical to the workspace agents OpenAI released the same day — deploying independent AI agents in the enterprise to reduce manual intervention in repetitive tasks.
Data sovereignty
Workspace Intelligence supports locking data to specific regions: the United States, the European Union, India, and separately controlled Germany.
This list isn't arbitrary. European customers' data compliance requirements have long been a pain point for Google Cloud, and the separate Germany control shows Google is taking this issue seriously. There's also an AI Control Center tool that lets administrators monitor and audit agent behavior logs.
Three players, all at once
April 22 was a dense day:
- OpenAI released workspace agents (Codex-driven, entering the enterprise from the ChatGPT portal).
- Google released Workspace Intelligence (Gemini-driven, entering the enterprise from the Gmail/Docs portal).
- Microsoft had earlier pushed Copilot Wave 3 (entering the enterprise from the Office 365 portal).
Each player has different cards: OpenAI has the broadest user base and the strongest model brand; Microsoft has deep embedding in Office and Teams; Google has 2 billion users' worth of Workspace data — the largest volume of emails, documents, and calendars.
But Google's biggest problem lies here too — the most data also means the biggest privacy concerns. Workspace Intelligence reads your Gmail to draft content. Whether users are willing to let it do that is the real variable determining how far this product can go.
Sources: Google's productivity suite gets AI smarts with Workspace Intelligence (SiliconANGLE); Google updates Workspace to make AI your new office intern (TechCrunch); CocoLoop; 10 more announcements for Workspace at Google Cloud Next 2026 (Google Workspace Blog); New Workspace Intelligence delivers unified, real-time understanding (Google Workspace Blog)